January 29-30 Oracle meltdown

compared with
Current by Tony Johnson
on Feb 01, 2008 09:56.


 
Key
These lines were removed. This word was removed.
These lines were added. This word was added.

View page history


There are 1 changes. View first change.

 We observed that the time taken to register datasets rose dramatically.
  
  !opssim2.png!
  !plot1.png!
  
 Possible causes:
  
 # Tom's was attempting to delete an old task with 1.5million datasets, starting at about 8am January 29. This request never completed and was canceled at about 5pm on January 29
 # Karen attempted to set up materialized views on the dataset table, at around 11am on January 29.
 # Glastlnx11 was importing large amounts of housekeeping data (exact time?)
 # There were problems with the nightly backup on January 29/30 night, it was restarted and ran for much of the day on January 30.